Let $X$ be an infinite-dimensional normed linear space. $\overline{B}(0,1):=\{x\in X:||x||\leq 1\}$. My question is, $\overline{B}(0,1)$ is totally bounded?

If $X$ is complete, that is to say, $X$ is a Banach space, then $\overline{B}(0,1)$ is closed, so $\overline{B}(0,1)$ is complete. But $\overline{B}(0,1)$ is not compact, so $\overline{B}(0,1)$ is not totally bounded.

If $X$ is not complete, then $\overline{B}(0,1)$ is totally bounded?

def: $(X,d)$ is a metric space, $A$ is a subset of $X$, $A$ is called totally bounded, if for any $\epsilon>0$, there exist a finite subset $F\in X$, such that $A\subseteq\bigcup\limits_{a\in F} B(a,\epsilon)$

Thanks very much for any help.


The answer is no for every infinite-dimensional normed linear space $X$. One can show this by employing Riesz's Lemma. We construct a sequence $(x_n)_{n=1}^\infty$ in the closed unit ball $B$ of $X$ such that $$\|x_m - x_n\| > 1/2,$$ if $m \neq n$. Then no finite collection of balls having radius $1/4$ can cover $B$ (for they can't cover $\{x_n\}_{n=1}^\infty$).

  • $\begingroup$ I know that the closed unit ball is not sequentially compact, but then B is not totally bounded? Why? $\endgroup$ – James Chan May 7 '16 at 13:24
  • $\begingroup$ Suppose $B$ could be covered by a finite collection of balls having radius $1/4$. Then there would be positive integers $m\neq n$ such that $x_m$ and $x_n$ are in the same ball (pigeonhole principle). By the triangle inaquality we would have $\|x_m−x_n\| \leqq 1/2$, which can't happen. So the definition fails for $ \epsilon=1/4$. It's not only that $(x_n)_{n=1}^\infty$ won't have a convergent subsequence, but each term is at distance $>1/2$ from each of the others. $\endgroup$ – user179514 May 7 '16 at 14:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.